The primary prevention of ovarian cancer through the identification of modifiable exposures or practical chemopreventive agents has been an elusive goal of epidemiologic studies of the disease, partly because the risk factors identified are relatively weak or inconsistent among studies. This project proposes two strategies to address these limitations. To explain why certain risk factors are modest in effect, it may be necessary to identify gene-environment interactions, modification between risk factors, or variation by histologic type of ovarian cancer. To address the issue of consistency, researchers with separate data sets should work more closely together during data analysis to be certain that common definitions of exposures and confounders are used. Consistency between case-control and cohort studies would provide especially compelling evidence about risk factors. For this proposal, researchers working with existing data and biologic specimens from two large studies, the Nurses' Health Study Cohort and a New England based case-control study of ovarian cancer, will join forces to address the following hypotheses. First, by free radical damage generated from chronic inflammation, genital exposure to talc increases risk for ovarian cancer; and the association may be modified by variants of genes in detoxification pathways. Second, by a variety of mechanisms, anti-inflammatory drugs may reduce the risk for ovarian cancer; and the association may be modified by variants of genes in metabolic pathways. Third, by enhancing steroid production, caffeine consumption may affect ovarian cancer risk differentially by menopausal status; and the association may be further modified by cholesterol consumption, smoking, hormonal use, or variants of genes in caffeine metabolism pathways. Finally, by their ability to scavenge reactive oxygen species or affect growth pathways, carotenoids may reduce risk for ovarian cancer; but a systematic investigation of factors contributing to heterogeneity between studies is necessary as well as an examination of the effect of other types of antioxidants such as the flavonoids. Histologic variation will be assessed by conducting analyses for all epithelial types combined and the following categories: serous invasive, serous borderline, mucinous, endometrioid and clear cell, and other/undifferentiated types. The common lifestyle exposures to be assessed in this project, as opposed to stronger risk factors like pregnancy history, may be more amenable to change. Clarifying their role would permit stronger public health advice and potentially decrease ovarian cancer occurrence.